It is widely known that radiation from the sun striking earth provides enough energy to supply all of mankind's needs for energy for the indefinite future. Such a source of energy can be clean and environmentally benign.
It is also widely known that global warming is associated with the use of fossil fuels, such as coal, oil, and natural gas. It is accepted by the scientific community that global warming can have severe adverse effects around the planet. There are numerous efforts around the world, combined with a sense of urgency, to cut down emissions from the usage of fossil fuels. A dominant factor in favor of the continual use of fossil fuels is their cost per unit of available energy. If, for example, the cost of producing photovoltaic cells can be reduced by a factor of approximately three while maintaining efficiency of conversion, the photovoltaic technology would become cost competitive with fossil fuels.
A major cost component in photovoltaic cells is the cost of the substrate on which the semiconductor film capable of converting sunlight into electricity is placed. The most widely used substrate is single crystal silicon (Si). These substrates developed for the microelectronics industry have been modified for application in photovoltaic technology. If a silicon film could be deposited on an inexpensive substrate, such as glass, and with comparable quality as that found in silicon single crystals used in the microelectronics industry, the cost of photovoltaic technology would drop significantly.
Epitaxial growth of thin films is a very well established process. It has been investigated by hundreds of researchers. Epitaxial deposition provides a very viable way of growing very good quality films. Many single crystal semiconductors and insulator surfaces are used to study the epitaxial growth of metallic films; for example, the growth of silver on silicon, sapphire, or a mica surface. Epitaxial metallic films have also been grown on other metallic films, such as gold on silver. In contrast to metals, semiconductors, such as silicon, are difficult to grow epitaxially. For example, heteroepitaxial films of silicon have been successfully grown only on sapphire but at temperatures that are relatively high for the applications we disclose here, such as the growth of silicon on glass substrates.
In order to take advantage of highly textured large grained films for photovoltaic technology two problems need to be solved: inexpensive growth of high quality films and the availability of an inexpensive substrate on which desirable properties can be achieved. Here, we disclose a method for growing semiconductor films, such as silicon, satisfying the two requirements listed above and suitable for photovoltaic technology and other electronic applications.
The thermodynamic stability and formation temperature of two or more elements is described by a composition versus temperature diagram, called a phase diagram. In this invention we shall make use of phase diagrams. These phase diagrams are available in the scientific literature (Massalski et al). The phase diagram provides information on the behavior of different phases, solid or liquid as a function of temperature and composition. For example, the liquidus in a simple binary eutectic system, such as Au and Si, shows how the relative composition of the liquid and solid, it is in equilibrium with, changes with temperature. It is therefore possible to choose an average composition, different from the eutectic composition, and cool the mixture in such a way as to precipitate out one phase or the other. If the composition is chosen to be richer in silicon than the eutectic composition then on cooling through the liquidus boundary between the single phase liquid and the two phase liquid plus solid, silicon will nucleate and form a solid phase. If on the other hand it is gold rich relative to the eutectic composition the first solid phase to nucleate is gold rather than silicon.
At and below the eutectic temperature the two components, in this case, Au and Si solidify from the liquid phase to phase separate into the two components Au and Si. The interface energy between the two components is generally positive and therefore drives the two components to aggregate into distinct phases with a minimum of surface area between the two rather than a fine mixture of the two. There is, however, the energetics of two other interfaces to consider also: one with the substrate and the other with vacuum or gas. In considering energetics it is not only the chemical interaction of the metal or Si with the substrate that is important but also its crystallographic orientation, for the surface or interface energy depends upon orientation of the grains. Another concern is the difference in lattice match between the nucleating film and the substrate which can lead to strain induced energy that is minimized by either inducing defects or not growing uniformly in thickness across the substrate surface. These factors determine if silicon is likely to deposit on the substrate (heterogeneous nucleation) or nucleate and forms small crystals in the liquid (homogeneous nucleation).
An advantage of using eutectics compositions is that the eutectic temperature is lower than the melting temperature of the constituent elements. For example, the eutectic temperatures of Au, Al, and Ag with Si are 363, 577, and 835 degrees Centigrade (° C.), respectively. In contrast the melting temperatures of the elements are 1064, 660, and 961° C., respectively. The melting temperature of silicon is 1414° C. The eutectics then offer the possibility of nucleating a silicon crystal from the liquid far below the temperature at which pure liquid silicon crystallizes. By a proper choice of the substrate surface exposed to the nucleating silicon, it is possible to nucleate and grow single crystal or large grained silicon films.
We have discussed silicon eutectics using elements such as Au, Ag, and Al. However, it is possible to replace the elements by silicon based compounds. For example, the compound nickel silicide forms a eutectic with Si. There are numerous other examples of silicide compounds forming a eutectic with Si (Massalski et al). An advantage of using a silicide is that frequently the electrical contact of the silicide with silicon has very desirable properties, such as a good ohmic contact or a Schottky barrier. Some silicides are also known to have an epitaxial relationship with silicon. In this case, by appropriately choosing either a silicide rich or silicon rich melt either the silicon can be induced to grow epitaxially on the silicide or the silicide on silicon. A disadvantage in this approach is the eutectic temperature, which is generally high.
Low temperature solutions can also be formed with some elements, For example, gallium (Ga) and Si have a eutectic temperature of less than 30° C., very close to that of the melting point of Ga. There are other elements, such as indium or tin that form low temperature liquid solutions with silicon. Si can be nucleated from these solutions at very low temperatures relative to pure silicon (Girault et al, Kass et al). These temperatures are sufficiently low that it opens up the possibility of using organic materials as substrates on which large grained to single crystal films can be grown. While this is an advantage, there is also a serious disadvantage; at these low temperatures, the silicon film can contain defects and hence are not very useful as a photovoltaic material. However, these very low temperature deposits can be used to initiate the nucleation of a very thin silicon film, which is subsequently thickened by using higher temperature processes to optimize its photovoltaic properties.
The choice of a particular system (phase diagram) is not only determined by temperature and energetics of the interfaces, but also by the solubility of the second element in Si. It is desirable to have precise control of the doping of Si in order to optimize its semiconductor properties for photovoltaic applications. It is also important to select the composition of the substrate and temperature of processing such that there is minimal or no chemical interaction between the silicon film and the surface of the substrate on which it is being deposited.
From the preceding description, we can extract five common points which are relevant to this invention. First, one end of the phase diagram always has the semiconductor we wish to nucleate and use to produce a film, we have used silicon in the preceding examples but it could be germanium or a compound such as gallium arsenide or cadmium selenide. Second, the thermodynamically predicted concentration of the second element or phase in the semiconductor is minimal. If there is solubility then it must be a desirable dopant. For example aluminum (Al) in silicon behaves as a p-type dopant and experience in the semiconductor industry has shown that trace amount of Al can be desirable. Third, the liquidus curve has the highest temperature on the semiconductor side. In other words, the melting point of the semiconductor is greater than the liquidus for all compositions in equilibrium with the semiconductor. Fourth, the homogeneous nucleation energy of silicon crystal from the melt is greater than that for heterogeneous nucleation on the substrate. This latter condition promotes heterogeneous nucleation. And, fifth, the temperature for epitaxial growth is low enough to use inexpensive substrates such as glass but high enough to promote a good quality silicon film. For example, a growth temperature above approximately 550 degrees Centigrade (550° C.) is desirable to make a good quality silicon film. The softening temperature of ordinary glasses is around 600° C. The softening temperature of borosilicate glasses is higher. However it is not high enough to use conventional deposition temperature of greater than 750 degrees Centigrade for silicon on insulator, such as a sapphire substrate.
In order to take full advantage of the invention disclosed here the semiconductor material has to be deposited on a substrate material which is inexpensive, and the surface of which enables heterogeneous nucleation and growth. In the following we shall discuss two specific methods for producing substrates suitable for heterogeneous deposition of films for photovoltaic technology. Both of these methods have been described in the scientific literature and we do not claim to invent them. We include them here for completeness.
The use of rolled and textured Ni and Ni-alloy sheets has been proposed as substrate material for superconducting films and, more recently, for films for photovoltaic devices (Findikoglu et al). In order to facilitate the growth of epitaxial superconducting films on such substrates, there have been two approaches described in the scientific literature: in one the sharp rolling texture produced in a rolled and annealed Ni alloy is used as a template on which various epitaxial buffer layers are deposited followed finally by an epitaxial film of a high temperature cuprate superconductors (Goyal et al). In the second approach (Findikoglu et al), the nickel ribbon is used as a substrate for ion beam assisted deposition of a wide variety of highly textured ceramics, for example, magnesium oxide (MgO). The ion beam aligns the growing MgO film, which provides a template for the subsequent deposition of the cuprate superconductor. The latter approach is not limited to using metal tapes but can be extended to other inexpensive substrates such as glass (Teplin et al). It has been found that texture can also be induced in MgO by depositing the film on a substrate that is inclined to the normal from the oncoming vapor of MgO.
One limitation of the use of glass as a substrate has been its softening temperature, which is generally lower than the conventional processing temperatures required for the growth of large grained or single crystal films of silicon. With the method of depositing silicon films at low temperatures, described in this invention, the use of buffered glass becomes an option for we can deposit highly textured and large grained silicon on MgO at or below the softening temperature of glass. Similarly, researchers have grown crystalline aluminum oxide (Al2O3) on inexpensive substrates (Findikoglu et al). We shall use MgO and Al2O3 as illustrative examples. However, it is understood to those skilled in the art that a variety of other materials can also work. Both Findikoglu et al and Goyal et al describe other buffer layers, including conducting ceramic layers, such as TiN.